So, you can still use HTTP, but it's probably overkill unless you want some sort of security.

Short of that, one solution is to set up a TCP server socket on the quake machine, and have it broadcast whatever's showing up to any and all clients that connect. Then write a simple TCP client on the perlbot side, and when data is ready, feed it to the perlbot.

-- Randal L. Schwartz, Perl hacker